Looping over column names like DF$j

Question

I have the following data frame:

    ID  A1  A2  A3    A4     B1      B2      B3      B4   
1  ID1   1   2   1     1      1       1       2       2
2  ID2   2   2   1     1      2      NA       2       1
3  ID3   2   2   1     2      1       1      NA       2          
4  ID4   1   1   1     1      1      NA      NA       2         
5  ID5   2   2   1     1     NA      NA       2      NA       
6  ID6   1   1   1     1      2       2       2       2

I want to extract rows from the data frame meeting specific conditions for each combination of A and B. These are the conditions:

PP: value 2 present in A-type column + value 2 present in B-type column
PA: value 2 present in A-type column + value 1 present in B-type column
AP: value 1 present in A-type column + value 2 present in B-type column
AA: value 1 present in A-type column + value 1 present in B-type column

This is what I wrote:

A <- colnames(dataframe)[2:5]
# A = A1, A2, A3 and A4
B <- colnames(dataframe)[6:9] 
# B = B1, B2, B3 and B4

for (a in A) {  
   for (b in B) {
         PP <- dataframe[dataframe$a=='2' & dataframe$b=='2' , ]
         PA <- dataframe[dataframe$a=='2' & dataframe$b=='1' , ]
         AP <- dataframe[dataframe$a=='1' & dataframe$b=='2' , ]
         AA <- dataframe[dataframe$a=='1' & dataframe$b=='1' , ]
        print(head(PP)) #to have a preview
    }
 }

However the new data frames are empty and I don't understand why. Ideally, the first for loop (with a=A1 and b=B1) would therefore output:

PP:

     ID  A1  A2  A3  A4  B1  B2  B3  B4
 2  ID2   2   2   1   1   2  NA   2   1

PA:

    ID  A1  A2  A3  A4  B1  B2  B3  B4
3  ID3   2   2   1   2   1   1  NA   2

AP:

    ID  A1  A2  A3  A4  B1  B2  B3  B4
6  ID6   1   1   1   1   2   2   2   2

AA:

    ID  A1  A2  A3  A4  B1  B2  B3  B4
1  ID1   1   2   1   1   1   1   2   2

I hope someone can help. Thanks.

Pierre L · Accepted Answer · 2016-09-07T13:41:52.247

1

The dollar sign extraction is creating a problem. It is not meant for passing of variables as in dataframe$a. Because a is not the actual name. You are trying to pass the variable name on to it. But that operator will look for the literal a column and not find it. Try dataframe[,a]

edited Sep 07 '16 at 13:41

answered Sep 07 '16 at 13:33

Pierre L

28,203
6
47
69

Looping over column names like DF$j

1 Answers1